摘要： 现有二维非局部均值降噪算法仅能抑制三维图像的层内噪声，无法利用层间信息对图像进一步降噪。针对该问题，分析印刷电路板在锥形束CT系统中所成图像的自相似性，将现有二维算法扩展到三维空间，提出基于分块处理的三维非局部均值降噪算法。实验结果表明，该算法可进一步抑制噪声，具有较高的计算效率。

关键词: 工业CT, 三维图像, 图像降噪, 分块, 非局部均值降噪

Abstract: Existing 2D non-local mean noise reduction algorithm can only suppress the noise inside the slice, it can not access the noise between slices. Aiming at this problem, this paper analyzes image self-similarity which is generated by Printed Circuit Board(PCB) in Cone Beam Computed Tomography(CBCT), extends existing 2D algorithm to 3D space, and proposes 3D non-local mean noise reduction algorithm based on partitioning processing. Experimental results show that this algorithm can further restrain noise, and it has higher computational efficiency.

Key words: Industry Computed Tomography(ICT), 3D image, image noise reduction, partitioning, non-local mean noise reduction

中图分类号: 

• TN911.73